mlsanityData to 7 Oct 2026

Unsupervised Learning of Temporal Abstractions with Slot-based Transformers

arXiv 2203.13573 ↗First posted 25 Mar 2022cs.LG cs.AI cs.NEcited by 14

Not on the map. Why?

Abstract (arXiv)

The discovery of reusable sub-routines simplifies decision-making and planning in complex reinforcement learning problems. Previous approaches propose to learn such temporal abstractions in a purely unsupervised fashion through observing state-action trajectories gathered from executing a policy. However, a current limitation is that they process each trajectory in an entirely sequential manner, which prevents them from revising earlier decisions about sub-routine boundary points in light of new incoming information. In this work we propose SloTTAr, a fully parallel approach that integrates sequence processing Transformers with a Slot Attention module and adaptive computation for learning about the number of such sub-routines in an unsupervised fashion. We demonstrate how SloTTAr is capable of outperforming strong baselines in terms of boundary point discovery, even for sequences containing variable amounts of sub-routines, while being up to 7x faster to train on existing benchmarks.

Key facts. Cited by 14 papers in the map (data through 7 Oct 2026). Builds on 36 of its 68 references that are in the map. First posted on arXiv 25 Mar 2022.

Builds on

36 of 68 references.

References · 36

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2014Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translationcited by 4,016
  4. 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,410
  5. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  6. 2021Decision Transformer: Reinforcement Learning via Sequence Modelingcited by 1,528
  7. 2021Perceiver: General Perception with Iterative Attentioncited by 1,089
  8. 2020Object-Centric Learning with Slot Attentioncited by 865
  9. 2018Diversity is All You Need: Learning Skills without a Reward Functioncited by 734
  10. 2019Compressive Transformers for Long-Range Sequence Modellingcited by 699
  11. 2021Offline Reinforcement Learning as One Big Sequence Modeling Problemcited by 636
  12. 2021Perceiver IO: A General Architecture for Structured Inputs & Outputscited by 616
24 more
  1. 2018Universal Transformerscited by 613
  2. 2016Adaptive Computation Time for Recurrent Neural Networkscited by 479
  3. 2016The Option-Critic Architecturecited by 433
  4. 2019Multi-Object Representation Learning with Iterative Variational Inferencecited by 359
  5. 2016Variational Intrinsic Controlcited by 262
  6. 2019Stabilizing Transformers for Reinforcement Learningcited by 262
  7. 2021Conditional Object-Centric Learning from Videocited by 247
  8. 2020On the Binding Problem in Artificial Neural Networkscited by 230
  9. 2016Modular Multitask Reinforcement Learning with Policy Sketchescited by 200
  10. 2020OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learningcited by 164
  11. 2021PonderNet: Learning to Pondercited by 156
  12. 2022TransDreamer: Reinforcement Learning with Transformer World Modelscited by 143
  13. 2019Reinforcement Learning Upside Down: Don't Predict Rewards -- Just Map Them to Actionscited by 124
  14. 2017Neural Expectation Maximizationcited by 123
  15. 2019Spatial Broadcast Decoder: A Simple Architecture for Learning Disentangled Representations in VAEscited by 116
  16. 2017A Laplacian Framework for Option Discovery in Reinforcement Learningcited by 108
  17. 2019Training Agents using Upside-Down Reinforcement Learningcited by 106
  18. 2019MCP: Learning Composable Hierarchical Control with Multiplicative Compositional Policiescited by 105
  19. 2021Going Beyond Linear Transformers with Recurrent Fast Weight Programmerscited by 84
  20. 2018TACO: Learning Task Decomposition via Temporal Alignment for Controlcited by 61
  21. 2020Attention over learned object embeddings enables complex visual reasoningcited by 59
  22. 2018Temporal Difference Variational Auto-Encodercited by 48
  23. 2019Variational Temporal Abstractioncited by 37
  24. 2021Learning Task Decomposition with Ordered Memory Policy Networkcited by 15

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 3; 2023: 5; 2024: 4; 2025: 0; 2026: 2.

14 papers.

Sort:

By area:

  1. 2023Object-Centric Slot Diffusioncited by 90
  2. 2023Rotating Features for Object Discoverycited by 29
  3. 2023Transformers in Reinforcement Learning: A Surveycited by 11
  4. 2024Neural Language of Thought Modelscited by 9
  5. 2024Recurrent Complex-Weighted Autoencoders for Unsupervised Object Discoverycited by 8
  6. 2023Learning Rational Subgoals from Demonstrations and Instructionscited by 5
  7. 2024Binding Dynamics in Rotating Featurescited by 5
  8. 2026Learning Abstractions for Hierarchical Planning in Program-Synthesis Agentscited by 3
  9. 2022Unsupervised Tokenization Learningcited by 2
  10. 2023Unsupervised Musical Object Discovery from Audiocited by 2
  11. 2024Parallelized Spatiotemporal Bindingcited by 2
  12. 2022Towards Improving the Generation Quality of Autoregressive Slot VAEscited by 1
2 more
  1. 2022Solving Reasoning Tasks with a Slot Transformercited by 1
  2. 2026Unsupervised Hierarchical Skill Discoverycited by 0